Intrusion Location with Breaking-the-Pulse-Width-Limit Spatial

Resolution and High Robustness based on φ-OTDR and

Spatial Frequency Analysis

Zhou Sha, Hao Feng, Yi Shi

and Zhoumo Zeng

State Key Laboratory of Precision Measurement Technology and Instruments, School of Precision Instrument and

Opto-electronic Engineering, Tianjin University, 92 Weijin Road, Tianjin 300072, China

Keywords: Phase-sensitive OTDR, Vibration Location, Spatial Resolution, Spatial-frequency, Energy Distribution.

Abstract: This paper proposes a vibration location method based on spatial-frequency analysis, which can help

achieve better location performance in the application of φ-OTDR. The method is proven to be able to

separately locate coupled vibration events to some extent, achieve break-the-pulse-width-limit spatial

resolution in the case of single event detection with higher robustness, and provide more comprehensive

information about the vibration situation to better identify vibration events, noise and disturbance. Different

experiments including indoor and outdoor tests are conducted to practically validate the effectiveness of the

method and its advantages over the conventional moving differential method. Experiment results show that

the proposed method exhibits several useful merits.

1 INTRODUCTION

Distributed optical fibre sensing technique based on

phase sensitive Optical Time Domain Reflectometry

(Taylor and Lee, 1993) has become increasingly

popular in recent years (Bao and Chen, 2012). With

the capability of simultaneous multipoint

monitoring, easiness of deployment and

maintenance, intrinsic immunity to chemical

corrosion and electromagnetic interference as well

as the cost effectiveness, φ-OTDR has been

extensively used in many fields such as intrusion

detection, perimeter security surveillance (Owen et

al., 2012), oil and gas pipeline monitoring (Peng et

al., 2014a), railway safety monitoring(Peng et al.,

2014b), structural health monitoring (Bahrampour et

al., 2010) as well as seismic applications. Many

researches have been conducted to improve the

sensing performance from different perspectives,

such as extending the sensing range from typically

tens of kilometres to over 100km (Wang et al.,

2014), raising the frequency response to break the

round-trip-time limit (He Q et al., 2015),

overcoming the nonlinear distortion of intensity

detection by conducting phase demodulation and so

forth (Wang et al., 2016).

Although the performance of φ-OTDR has been

greatly improved from many perspectives (Wu et al.

2015), the research pertaining to enhancing the

capability of vibration location is rather limited.

Since having the vibration events located is

generally fundamental and prior to other subsequent

analysis, it is necessary to spend efforts improving

the location capability. Traditional ways of locating

intrusions include moving differential (Lu et al.,

2010) and distributed parameter computation (Shi et

al., 2016), which both have their noticeable

drawbacks. The working principle of moving

differential is based on the time signal fluctuation. It

picks out those positions with high fluctuation

amplitude as the results of intrusion location. Since

the laser frequency drift and environmental

temperature change can also cause signal

fluctuations, fake differential peaks and false

location results can be generated when moving

differential is used in practical applications. The

distributed parameter computation computes some

parameter of the time signal in each position as an

indicator of the presence of intrusion. The parameter

can be selected from the perspective of statistics and

information theory, such as entropy and correlation

dimension (Shi et al., 2016). Although this method

generally gives a better location performance

Sha, Z., Feng, H., Shi, Y. and Zeng, Z.

Intrusion Location with Breaking-the-Pulse-Width-Limit Spatial Resolution and High Robustness based on ϕ-OTDR and Spatial Frequency Analysis.

DOI: 10.5220/0006383700170023

In Proceedings of the 14th International Joint Conference on e-Business and Telecommunications (ICETE 2017) - Volume 3: OPTICS, pages 17-23

ISBN: 978-989-758-258-5

compared to moving differential, the computation

load is somehow raised in a significant level,

regardless of which specific indicator is chosen. As a

consequence, real time monitoring is unachievable

and the method can hardly be applied in practical

applications. On the other hand, the spatial

resolution is limited by the probe pulse width.

Higher pulse width would lead to lower spatial

resolution, thus deteriorates the location accuracy.

Our work proposed a new vibration location

method using spatial frequency analysis. It reports

the vibration condition in the entire sensing range in

a straightforward and intuitive way. The energy

distribution along the fibre of each frequency

component and the power spectrum at each position

are simultaneously reflected in one colour-encoded

figure. Since vibrations in different frequency bands

are independently presented at different positions

along the frequency axis, fake fluctuations caused by

laser frequency drift and environmental temperature

change, which more often appears with lower

frequency, can be conveniently distinguished and

thus will not lead to confusion. Besides, thanks to

the Fast Fourier Transform algorithm, this method

based on spatial frequency analysis is high-

efficiency and time-saving. In addition, combined

with the results of Shi et al., (2015), this method can

explicitly determine the edge of the vibration

segment, thus realize break-the-pulse-width-limit

spatial resolution for single event detection with

higher robustness. For overlapped multiple vibration

events with unique power spectrum, this method can

also help separate the events to a certain extent, and

thereby achieve higher general spatial resolution.

2 METHODOLOGY

The principle of the proposed spatial-frequency

analysis is illustrated in Fig. 1. The blue matrix is

essentially composed of multiple time-aligned

Rayleigh backscattered (RBS) traces, with the

horizontal axis being spatial location and the vertical

axis being time. Performing FFT transform

vertically to the blue matrix yields the red matrix,

which presents the energy distribution with respect

to space and frequency. Different columns of the red

matrix represent the power spectrum density at

different positions. Different rows of the red matrix

represent the energy distribution of different

frequency components along the fibre. As can be

seen, with the help of this frequency-spatial energy

distribution matrix, one can easily and intuitively

obtain the frequency information at each location.

The position without vibration would exhibit low-

level and uniform energy distribution, yet the

position affected by external vibration would exhibit

unique power distribution pattern, which is

essentially the power spectrum. Thus, vibration

events can be easily located by inspecting the energy

distribution simultaneously in the spatial and

frequency domain.

Figure 1: Principle illustration.

Assume the length of the sensing fibre is L, the

sampling rate of the data acquisition card is f, the

pulse repetition rate is N, the refractive index of

fibre is n, and the raw data matrix (blue matrix) is

made up of data generated within one second, then

the raw data matrix can be expressed as



ITT TT 

(1)

i1 2

[S S S ]

iiiM





(2)

nfL c



(3)

, which is made up of M sample points, denotes

the ith RBS trace. By concurrently performing FFT

to each column of I, the spatial-frequency energy

distribution matrix is generated and expressed as





OFF FF 

(4)

1j 2j

[A A A ]

F  

(5)





is the power spectrum of time signal 



. The value

of 



is calculated by means of FFT algorithm,

i2 /

ik i i N

ij kj N N

ASwwe







 



(6)

where i is the imaginary unit, i is the discrete

frequency index. This allows us to obtain the spatial

OPTICS 2017 - 8th International Conference on Optical Communication Systems

distribution of each and every discrete frequency

component i ,

[ ]

iii iM

DAA A 

(7)





denotes the spatial energy distribution of discrete

frequency i.

The method is essentially mapping the signal

energy into a spatial-frequency plane from which

one can independently investigate each discrete

frequency’s distribution along the fibre and the

unique power spectrum of time signal in each

position. One such colour-encoded plane provides

the general view of the entire sensing range in one

second. Positions with vibration can be distinctly

revealed, and its specific power spectrum is at the

same time readable. Low frequency fluctuation

caused by laser drift and temperature change will

only appear in the low frequency band, and the

positions with real vibration will have unique energy

distribution in the high frequency band. Thus the

fake vibration can be distinguished from the real

ones by looking at the spatial-frequency energy

distribution pattern and inspecting the frequency

band they occupy.

3 EXPERIMENTAL

VALIDATION

In order to validate the effectiveness of the method,

different experiments including indoor and outdoor

vibration tests are conducted where the vibration

events are located using the proposed spatial

frequency analysis. The system setup is shown in

Fig. 2. An NKT ultra narrow line width laser is used

as the light source. The continuous seed light is

modulated into pulses with an Acoustic Optic

Modulator (AOM). After that, the pulses are boosted

by an Erbium Doped Fibre Amplifier (EDFA) to

obtain higher peak power. A Fibre Bragg Grating

(FBG) is employed together with an optical

circulator to filter out the Amplifier Spontaneous

Emission (ASE) noise from EDFA. A second AOM

is applied right after the FBG to enhance the

Extinction Ratio (ER). The pulsed probe light is

injected into the sensing fibre through an optical

circulator and the Rayleigh Back Scattered light

(RBS) is detected with a PIN detector. The signal is

sampled with a 50MS/s Data Acquisition Card

(DAC) and processed subsequently in a computer.

Three types of vibrations, respectively caused by

Piezoelectric Transducer cylinder (PZT tube),

striking the steel plate and water pipe leakage are

measured with the φ-OTDR system.

Figure 2: Experimental setup φ-OTDR structure.

3.1 PZT Vibration Test

Firstly, two PZT tubes are used to apply sinusoidal

vibration to the fibre. A function generator drives the

two PZT tubes with the corresponding frequency

being 170Hz and 70Hz. The fibre’s length between

the two PZT tubes is 6m. A schematic diagram of

the deployment structure is illustrated in Fig. 3.

Figure 3: PZT vibration setup.

The vibration caused by the two PZT tubes are

firstly located using the conventional moving

differential method. The location result is shown in

Fig. 4.

Figure 4: Moving differential traces.

It can be observed from Fig. 4 and the inset that a

differential peak appears at around 160 m, which

indicates the presence of vibration event at the

Intrusion Location with Breaking-the-Pulse-Width-Limit Spatial Resolution and High Robustness based on ϕ-OTDR and Spatial Frequency

Analysis

mentioned location. The 160m location result is

basically correct, however, the starting point of the

vibration segment is not sufficiently clear. The two

individual vibration events respectively caused by

the two PZT tubes with different frequency can not

be distinguished. This partly exhibits the intrinsic

drawbacks of the moving differential locating

algorithm.

As comparison, the proposed spatial frequency

analysis is performed on the same set of experiment

data, and the location result is shown in Fig. 5 and

Fig. 6, which are essentially colour-encoded energy

distribution matrix, describing how the signal energy

is distributed with respect to space and frequency.

From Fig. 5 it can be observed that there is an

concentrated energy distribution around the position

160m. The corresponding frequency range at that

position approximately starts from 0Hz and ends at

1kHz. Zooming in Fig. 5 gives us Fig. 6, which

clearly exhibits the frequency distribution pattern of

the vibration in detail. Based on Fig. 6 we are

allowed to infer the presence of two individual

vibration events.

Figure 5: Spatial frequency energy distribution.

Figure 6: Spatial frequency energy distribution (detail).

The three thick lines horizontally starting from

150m, vertically appearing at 170Hz, 340Hz and

510Hz are all caused by the 170Hz PZT vibration,

where the 340Hz line and 510Hz line appear as a

result of the interference induced nonlinear

distortion of intensity signal. The four lines

horizontally ending at about 175m, vertically

appearing at 70Hz, 140Hz, 210Hz and 280Hz are

caused by the 70Hz PZT. Similarly, the three lines

apart from the 70Hz line are present due to the

interference induced nonlinear distortion, which can

be mathematically explained by the Jocobi-Angry

expansion. Therefore, the processing result of the

proposed spatial frequency analysis is able to

provide a much larger volume of information than

moving differential can do. It helps distinguish

coupled vibration events based on their unique

frequency distribution patterns. Besides, by

synthetically taking into account the emergence

position of each frequency component, one can

determine the staring point of the vibration segment

with higher robustness. It is revealed from the

research results in [3] that the starting point of the

vibration segment on the intensity trace is the real

indication of the vibration location. In the case of

single event detection, distinctly determining the

starting point of vibration segment means one can

achieve spatial resolution that breaks the pulse width

limit. Therefore, the proposed method can also help

locate single vibration event with higher accuracy

and spatial resolution that is decoupled with the

probe pulse width.

Based on the results given in Fig. 6, we can infer

that 170Hz and 70Hz are the two characteristic

frequency components that can be used to reveal the

respective positions of the two vibration events.

Thus the energy distribution along the fibre of the

two characteristic frequency components are

extracted and plotted in Fig. 7. Two peaks with

rather high signal to noise ratio are clearly present at

around 160m. From the inset, it can be seen the two

peaks respectively start from 151m and 157m. The

6m interval agrees well with the fibre length

between the two PZT tubes.

Figure 7: Spatial distribution of characteristic frequencies.

3.2 Steel Plate Striking Test

For the purpose of testing the location performance

OPTICS 2017 - 8th International Conference on Optical Communication Systems

of the proposed method in practical vibration

condition, the second experiment is designed where

a steel plate struck by a hammer is used as the

vibration source. The deployment schematic

diagram is shown in Fig. 8. The fibre is coupled with

the steel plate by vacuum coupling agent which

assures better plate-fibre vibration transmission. The

steel vibration signal is monitored using the home

made φ-OTDR system as shown in Fig. 2 and is

firstly located using the conventional moving

differential method with the result shown in Fig. 9.

A peak can be found at around 155m on the

differential trace, which indicates the presence of

vibration event. As can be seen from the inset, the

peak’s starting point is not clear and the signal to

noise ratio is rather low, thus the vibration location

can not be accurately determined.

The same experiment data is then processed

using the proposed spatial frequency analysis and

the location result is shown in Fig. 10. A vertical

line at around 150m is explicitly shown. Having the

line and its vicinity zoomed in and shown in the

inset, the starting position can be accurately

determined to be 150m. Besides, the frequency

component of the vibration is at the same time

revealed. The energy is shown to cover the entire

frequency response range, i.e., 0-2.5kHz.

Figure 8: Striking steel plate.

Based on the information given in Fig. 10, the

power spectrum of the signal at 150m is plotted to

further investigate the energy distribution, which is

shown in Fig. 11. Four maximum frequency

components, respectively 93Hz, 204Hz, 387Hz and

705Hz are picked out as the characteristic

frequencies of the vibration event. The energy

distribution of the four characteristic frequencies

along the sensing fibre are plotted in Fig. 12. It can

be observed that the energy level of each of the four

frequency components reaches its maximum value

within the range of 150m to 160m. Therefore, the

spatial energy distribution of each of the four

characteristic frequencies can be used to determine

the vibration location.

Figure 9: Moving differential traces.

Figure 10: Spatial frequency energy distribution.

This part of the experiment results show that the

proposed spatial frequency analysis can provide

more abundant information to help locate and

identify the vibration event, such as revealing the

frequency spectrum, indicating the characteristic

frequency, explicitly and robustly give the start

position of the vibration segment.

Figure 11: Characteristic frequencies.

Figure 12: Spatial distribution of characteristic frequencies.

Intrusion Location with Breaking-the-Pulse-Width-Limit Spatial Resolution and High Robustness based on ϕ-OTDR and Spatial Frequency

Analysis

3.3 Pipe Leakage Vibration Test

Pipeline integrity monitoring is one of the major

applications of φ-OTDR, where the detection and

location of external intrusion and pipeline accident

such as pipe leakage is necessary. An experiment

pertaining to pipeline leakage location is designed

using the home made φ-OTDR system and

conducted inside a greenhouse. The field setup is

schematically shown in top view in Fig. 13. A PE

water pipe is buried under ground, with about 15m

of it immerged in a dug water tank. An 8mm hole is

drilled on the pipe shell to simulate pipe leakage. A

water pump is used to pump the water from a water

well into the water pipe. The sensing fibre is glued

on the pipe shell surface. When the water pump

comes into operation, water leakage would occur

through the leakage hole, and the leakage induced

vibration is monitored by the sensing fibre.

Figure 13: Pipeline leakage experimental setup.

The leakage event is respectively located using

the conventional moving differential method and the

proposed spatial frequency analysis. The processing

result of moving differential is presented in Fig. 14

and Fig. 15. A conspicuous peak appears at around

650m in Fig. 14, which basically gives the correct

position of the leakage event. However in Fig. 14

where the differential peak is zoomed in for clearer

observation, it is hard to tell the starting point of the

vibration segment in that an explicit bound between

the peak and the background trace can hardly be

found. Therefore, the location accuracy is rather

limited.

By performing the proposed spatial frequency

analysis, the leakage event could be better located

with higher accuracy. Fig. 16 gives the spatial

frequency energy distribution in a full view. The

energy line at about 650m reveals the correct

leakage position. In Fig. 17 the energy line is

zoomed in where the start position of vibration

segment can be explicit determined. It can be

observed according to the part of line vertically from

300Hz to 500Hz that the precise start point is 657m.

Disturbance are present in the frequency range

below 200Hz, mainly in 50Hz, 100Hz and 150Hz.

This noise is validated to be caused by the water

pump. When the water pump engine is turned on, the

disturbance accordingly comes about. When the

water pump engine is turned off, the disturbance

disappears.

Figure 14: Moving differential traces.

Figure 15: Moving differential traces (detail).

In general, based on the spatial frequency energy

distribution, one can clearly observe the presence of

the disturbance caused by the water pump engine

vibration, accurately determine the leakage position

by referring to the part of energy distribution over

200Hz and even identify those positions that are

contaminated by the water pump engine noise.

Spatial frequency energy distribution provides more

useful and comprehensive information about the

vibration situation in the monitoring field and thus

helps achieve better location performance.

Figure 16: Spatial frequency energy distribution.

OPTICS 2017 - 8th International Conference on Optical Communication Systems

Figure 17: Spatial frequency energy distribution (detail).

4 CONCLUSIONS

A vibration location method used in φ-OTDR

system based on spatial frequency analysis is

proposed and demonstrated. Different experiments

including indoor and outdoor tests are conducted to

comprehensively evaluate the method’s locating

performance in different situations. The method has

been proven to exhibit several useful advantages

over the conventional moving differential method.

Firstly, it provides vibration information in a more

comprehensive way. By referring to the vibration

energy distribution with respect to space and

frequency, one can easily achieve an intuitive

comprehension about the vibration condition in the

entire sensing range. The colour-encoded image can

conveniently reflect how the energy is distributed in

the frequency domain in each position and how the

energy of each frequency component is distributed

along the fibre. Vibration caused by different

sources can be separated according to their unique

frequency spectrums. By synthetically considering

the emerging position of each frequency component,

one can determine the start position of vibration

segment in a more accurate and robust way, which

will lead to better location results. According to the

research result reported in (Shi et al., 2015), it is the

start point of the vibration segment that reveals the

real vibration location, thus accurately determining

the border of vibration segment is of great

importance in that the pulse width limit upon the

spatial resolution can be broken through. Therefore

this method can help achieve break-the-pulse-width-

limit spatial resolution with higher robustness.

Apart from the location information, the

frequency spectrum of each detected vibration is

reflected at the same time. Since vibration caused by

different sources generally exhibit different

characteristic in the frequency domain, this property

is thus very useful to help identify different

vibrations.

Besides, thanks to the mature application of the

FFT algorithm, the proposed method can be

performed in a very convenient and time-saving way.

Real time monitoring based on the proposed method

is generally available.

ACKNOWLEDGEMENTS

The authors appreciate the financial support from the

National Natural Science Foundation of China (No.

61304244), the Natural Science Foundation of

Tianjin (14JCQNJC04900), the Research

Foundation for the Doctoral Program of Higher

Education of China (20130032120066).

REFERENCES

H. F. Taylor and C. E. Lee, 1993. Apparatus and method

for fibre optic intrusion sensing. U.S. Patent 5194847.

X. Bao and L. Chen, 2012. Recent progress in distributed

fibre optic sensors. Sensors (Basel) 12(12), 8601–

8639.

Shi. Y et al., 2015. A long distance phase-sensitive optical

time domain reflectometer with simple structure and

high locating accuracy. Sensors (Basel) 15(9), 21957–

21970.

Wu. H et al., 2015. Separation and determination of the

disturbing signals in phase-sensitive optical time

domain reflectometry (Ф-OTDR). J. Lightwave

Technol. 33(15), 3156–3162.

Peng. F et al., 2014a. Ultra-long high-sensitivity Φ-OTDR

for high spatial resolution intrusion detection of

pipelines. Opt. Express 22(11), 13804–13810.

Peng. F et al., 2014b. Real-time position and speed

monitoring of trains using phase-sensitive OTDR.

IEEE Photonics Technol. Lett. 26(20), 2055–2057.

Owen, G. Duckworth and J. Worsley, 2012. Fibre optical

distributed acoustic sensing for border monitoring.

2012 European Intelligence and Security Informatics

Conference, 362–364.

Shi. Y et al., 2016. Correlation dimension locating method

for phase-sensitive optical time domain reflectometry.

Optical Engineering, 55(9): 091402-091402.

Bahrampour A R et al., 2010. Resolution enhancement in

long pulse OTDR for application in structural health

monitoring. Optical fibre technology, 16(4): 240-249.

Wang. Z et al., 2014. Ultra-long phase-sensitive OTDR

with hybrid distributed amplification. Optics letters,

39(20): 5866-5869.

He. Q et al., 2015. Frequency response enhancement by

periodical non-uniform sampling in distributed

sensing. IEEE Photonics Technology Letters, 27(20):

2158-2161.

Wang. Z et al., 2016. Coherent Φ-OTDR based on I/Q

demodulation and homodyne detection. Optics

express, 24(2): 853-858.

Intrusion Location with Breaking-the-Pulse-Width-Limit Spatial Resolution and High Robustness based on ϕ-OTDR and Spatial Frequency

Analysis